FPGA-Based High-Throughput CNN Hardware Accelerator With High Computing Resource Utilization Ratio.

The field-programmable gate array (FPGA)-based CNN hardware accelerator adopting single-computing-engine (CE) architecture or multi-CE architecture has attracted great attention in recent years. The actual throughput of the accelerator is also getting higher and higher but is still far below the theoretical throughput due to the inefficient computing resource mapping mechanism and data supply problem, and so on. To solve these problems, a novel composite hardware CNN accelerator architecture is proposed in this article. To perform the convolution layer (CL) efficiently, a novel multiCE architecture based on a row-level pipelined streaming strategy is proposed. For each CE, an optimized mapping mechanism is proposed to improve its computing resource utilization ratio and an efficient data system with continuous data supply is designed to avoid the idle state of the CE. Besides, to relieve the off-chip bandwidth stress, a weight data allocation strategy is proposed. To perform the fully connected layer (FCL), a single-CE architecture based on a batch-based computing method is proposed. Based on these design methods and strategies, visual geometry group network-16 (VGG-16) and ResNet-101 are both implemented on the XC7VX980T FPGA platform. The VGG-16 accelerator consumed 3395 multipliers and got the throughput of 1 TOPS at 150 MHz, that is, about 98.15% of the theoretical throughput ( 2 ×3395 ×150 MOPS). Similarly, the ResNet-101 accelerator achieved 600 GOPS at 100 MHz, about 96.12% of the theoretical throughput ( 2 ×3121 ×100 MOPS).

Association between switching of primary outcomes and reported trial findings among randomized drug trials from China.

BACKGROUND AND OBJECTIVE: To evaluate the association between the nature of findings and the switching of registered primary outcomes among randomized controlled trials (RCTs) from mainland China. METHODS: This is a retrospective cohort study. We retrieved RCTs from trial registries and identified the corresponding journal articles from bibliographic databases until August 2019. Trial registries and journal articles were compared to evaluate whether registered primary outcomes with negative findings were more likely to be switched to secondary outcomes in the subsequent journal articles than those with positive findings. RESULTS: Switching of registered primary outcomes occurred in 131 (45%) of 294 RCTs. A total of 450 registered primary outcomes were matched to 522 (37%) primary outcomes and 871 (63%) secondary outcomes in the journal articles. Among RCTs registered before they started, the odds of switching primary outcomes with negative findings were 2.64 (95% CI: 1.16-6.02) times the odds of switching those with positive findings. Among RCTs registered when they were ongoing, the odds of switching primary outcomes with negative findings were 8.84 (95% CI: 3.62-25.93) times the odds of switching those with positive findings. CONCLUSION: The nature of findings may play a role in how likely a prespecified primary outcome is switched subsequently.

Assessment of Duplicate Publication of Chinese-Sponsored Randomized Clinical Trials.

Importance: Duplicate publications of randomized clinical trials are prevalent in the health-related literature. To date, few studies have assessed the interaction between duplicate publication and the language of the original publication. Objective: To assess the existence of duplicate publication and the extent to which duplicate publication is associated with the language of the original publication. Design, Setting, and Participants: In this retrospective cohort study, eligible randomized clinical trials were retrieved from trial registries, and bibliographic databases were searched to determine their publication status. Eligible randomized clinical trials were for drug interventions from January 1, 2008, to December 31, 2014. The search and analysis were conducted from March 1 to August 31, 2019. The trial registries were either primary registries recognized by the World Health Organization or the Drug Clinical Trial Registry Platform sponsored by the China Food and Drug Administration. Exposures: Individual randomized clinical trials with positive vs negative results. Main Outcomes and Measures: Journal articles were classified as main articles (determined by largest sample size and longest follow-up among all journal articles derived from that randomized clinical trial) and duplicates. The duplicates were classified into 4 types: (1) unreferenced subgroup analysis (article did not disclose itself as a subgroup analysis or reference its main article); (2) unreferenced republication (article did not disclose itself as a replicate of the main article or reference it); (3) unreferenced interim analysis (article did not disclose itself as an interim analysis or reference its main article); and (4) partial duplicate (article did not disclose its sharing a subset of participants with other articles or reference them). Results: Among 470 randomized clinical trials published by August 2019 as journal articles, 55 (11.7%) had 75 duplicates, of which 53 (70.7%) were cross-language duplicates. Of the 75 duplicates, 33 (44.0%) were unreferenced republications, 25 (33.3%) unreferenced subgroup analyses, 15 (20.0%) unreferenced interim analyses, and 2 (2.7%) partial duplicates. When the main article of a randomized clinical trial was published in Chinese, those with positive findings were 2.48 (95% CI, 1.08-5.71) times more likely to have subsequent duplicate publication than those with negative findings. Conclusions and Relevance: In this study, most duplicates were cross-language duplicates and the most common type was unreferenced republication of the main article. Duplicate publication bias exists when the main articles of randomized clinical trials were published in Chinese, potentially misleading readers and compromising journals and evidence synthesis.

Assessment of Language and Indexing Biases Among Chinese-Sponsored Randomized Clinical Trials.

Importance: Language and indexing biases may exist among Chinese-sponsored randomized clinical trials (CS-RCTs). Such biases may threaten the validity of systematic reviews. Objective: To evaluate the existence of language and indexing biases among CS-RCTs on drug interventions. Design, Setting, and Participants: In this retrospective cohort study, eligible CS-RCTs were retrieved from trial registries, and bibliographic databases were searched to determine their publication status. Eligible CS-RCTs were for drug interventions conducted from January 1, 2008, to December 31, 2014. The search and analysis were conducted from March 1 to August 31, 2019. Primary trial registries were recognized by the World Health Organization and the Drug Clinical Trial Registry Platform sponsored by the China Food and Drug Administration. Exposures: Individual CS-RCTs with positive vs negative results (positive vs negative CS-RCTs). Main Outcomes and Measures: For assessing language bias, the main outcome was the language of the journal in which CS-RCTs were published (English vs Chinese). For indexing bias, the main outcome was the language of the bibliographic database where the CS-RCTs were indexed (English vs Chinese). Results: The search identified 891 eligible CS-RCTs. Four hundred seventy CS-RCTs were published by August 31, 2019, of which 368 (78.3%) were published in English. Among CS-RCTs registered in the Chinese Clinical Trial Registry (ChiCTR), positive CS-RCTs were 3.92 (95% CI, 2.20-7.00) times more likely to be published in English than negative CS-RCTs; among CS-RCTs in English-language registries, positive CS-RCTs were 3.22 (95% CI, 1.34-7.78) times more likely to be published in English than negative CS-RCTs. These findings suggest the existence of language bias. Among CS-RCTs registered in ChiCTR, positive CS-RCTs were 2.89 (95% CI, 1.55-5.40) times more likely to be indexed in English bibliographic databases than negative CS-RCTs; among CS-RCTs in English-language registries, positive CS-RCTs were 2.19 (95% CI, 0.82-5.82) times more likely to be indexed in English bibliographic databases than negative CS-RCTs. These findings support the existence of indexing bias. Conclusions and Relevance: This study suggests the existence of language and indexing biases among registered CS-RCTs on drug interventions. These biases may distort evidence synthesis toward more positive results of drug interventions.

Analysis and Measurement of China's Population Health Informatization Development Strategy.

This paper analyzes the development strategy of population health informatization in China, and summarizes the measurement direction and evaluation elements of population health informatization. Methods: Literature and field investigation, expert consultation and PEST analysis were used to determine the development level measurement and evaluation framework. Results: Based on the PEST analysis framework, the development level measurement and evaluation factors were defined, and the evaluation framework was established, which included system construction, IT application, information financing, information personnel, information policy and management, and information application effect. The information from hospitals and grass- roots medical and health institutions was also provided. From the perspective of the level of development, the framework of informatization evaluation is further refined.

Aislamiento y caracterización de una bacteria marina degradadora de testosterona: Vibrio sp. N3 / Isolation and characterization of a marine testosterone-degrading bacterium: Vibrio sp. N3

Steroids, including testosterone, estrone, 17β-estradiol, estriol and 17β-ethinyl estradiol, are harmful not only to the population dynamics of aquatic life forms but also to public health. In this study, a marine testosterone-degrading bacterium (strain N3) was isolated from Nanao Island in the South China Sea. In addition, the strain could also use 17β-estradiol (E2), 17β-ethinyl estradiol (EE2), estriol (E3) or cholesterol as a sole carbon source. According to the 16S rRNA gene sequence analysis, strain N3 was identified as Vibrio sp. Further characterization showed that the strain is aerobic, gram-negative, and mobile and exhibits resistance to ampicillin, carbenicillin, penicillin and spectinomycin. For enhancing its capacity of testosterone degradation, the Plackett-Burman factorial design and the central composite design were used to optimize the culture condition. Under optimal conditions, 92% of testosterone was degraded by Vibrio sp. N3 in 48 h.

Los esferoides-que incluyen la testosterona, la estrona, el 17 β-estradiol, el estriol y el 17 p-etinilestradiol-son nocivos no solo para la población dinámica de las formas de vida acuática, sino también para la salud pública. En este estudio se aisló una bacteria marina degradadora de testosterona de la isla de Nanao, en el Mar del Sur de China, a la que se denominó cepa N3. Se determinó que esta cepa también podría usar 17 β-estradiol (E2), 17 p-etinilestradiol (EE2), estriol (E3) o colesterol como únicas fuentes de carbono. De acuerdo con el análisis de la secuencia del gen 16S rRNA, la cepa N3 se identificó como Vibrio sp. La caracterización adicional mostró que dicha bacteria es un organismo aerobio, gram negativo y móvil, y que presenta resistencia a ampicilina, carbenicilina, penicilina y espectinomicina. Para optimizar la condición de cultivo en relación con su capacidad de degradar la testosterona, se utilizaron el diseño factorial Plackett-Burman y el diseno compuesto central. En condiciones óptimas, el 92% de la testosterona fue degradada por Vibrio sp. N3 en 48 h.

Isolation and characterization of a marine testosterone-degrading bacterium: Vibrio sp. N3.

Steroids, including testosterone, estrone, 17ß-estradiol, estriol and 17ß-ethinyl estradiol, are harmful not only to the population dynamics of aquatic life forms but also to public health. In this study, a marine testosterone-degrading bacterium (strain N3) was isolated from Nanao Island in the South China Sea. In addition, the strain could also use 17ß-estradiol (E2), 17ß-ethinyl estradiol (EE2), estriol (E3) or cholesterol as a sole carbon source. According to the 16S rRNA gene sequence analysis, strain N3 was identified as Vibrio sp. Further characterization showed that the strain is aerobic, gram-negative, and mobile and exhibits resistance to ampicillin, carbenicillin, penicillin and spectinomycin. For enhancing its capacity of testosterone degradation, the Plackett-Burman factorial design and the central composite design were used to optimize the culture condition. Under optimal conditions, 92% of testosterone was degraded by Vibrio sp. N3 in 48h.

Developing a Basic Medical Insurance Statement in Chinese Underdeveloped Areas.

BACKGROUND: We presented the running state of rural basic medical insurance system in Henan and discussed the enforcements and development experiences of underdeveloped areas. We provided data evidence to support the improvement and development of a basic rural medical insurance system. METHODS: We selected Henan Province, China as a sample, using the method of cluster sampling, from policy documents published in the national and provincial level of the new rural cooperative medical policy and work documents, data from 2004 to 2014, the National Health Statistical Yearbook of health statistics yearbook of Henan Province and relevant statistical data of the province. RESULTS: The new rural cooperative policy has covered the whole population in Henan Province. The number of individual received benefits is increasing. In 2013, the number of persons counted has reached to 270 million, funds raised and expenditures reached 38.5 billion and 26 billion, respectively. The operational task force has been developed rapidly. In 2013, on average each staff managed the cases for 16.4 thousand rural residents. CONCLUSION: The major implementation and development experience from the new rural cooperative policy of Henan province include: education of related knowledge, optimization of compensation plan, development of operational system and framework, improvement of management rules, reinforcement of information system development and financial supervision and increment of investment in rural medical healthcare.

[Cost analysis of the colorectal neoplasm screen program in Beijing].

OBJECTIVE: To conduct with a cost analysis of the colorectal neoplasm screening program in Beijing, and provide data evidence for decision making. METHODS: Based on stratified cluster sampling method, we carried out a 2-stage colorectal neoplasm screening program within 6 districts, Dongcheng, Xicheng, Chaoyang, Haidian, Fengtai and Shijingshan, of Beijing city between October, 2012 to May. 2013. The first stage of the program was to conducting a cancer risk level evaluation for community residents who were forty years older and the second stage's task was to providing clinical exam for those high risk people who were selected from the first stage. There were about 12 953 residents were involved in this program. We calculated the main cost of the colorectal neoplasm screen program in Beijing. Then estimate the cost of detecting one Colorectal Neoplasm patient of this program and compare it with the total treatment cost for a patient. RESULT: 2 487 high risk residents were selected by the first stage and 1 055 of them made appointment for the colonoscopy exam but only 375 accepted the exam, participate rate was 35.5%. 9 neoplasm cancer patients and 71 pre-cancer patient were found at the second stage, the detection rate were 69.2/100 000 and 546/100 000, respectively. The direct input for this neoplasm screening program was 227 100 CNY and the transport expense was 4 200 CNY in the calculations. The cost for detecting one cancer patient was about 19 900 CNY. Comparing with the total medical care cost of a cancer patient (1 282 800 CNY), especially for those have been diagnosed as middle to end stage cancer, the screening program (cost 842 800 CNY) might help to reduce the total health expenditure about 128 700 CNY, based on 12 953 local residents age above 40 years old. CONCLUSION: An colonoscopy based colorectal neoplasm screening program showed its function on medical expenditure saving and might have advantage on health social labor creating.